FreeBSD 11.0 Alpha 4 Released

[nasyt]

I know, but Michael used to troll a lot against Linux in the past when its drivers were far from perfect. That's why I wonder why he's so big-hearted when comes to bsd?

He's not being big hearted, your anti-BSD worldview is simply coloring intention into his not benchmarking BSD's graphics. Michael basically only benchmarks the BSDs at all when they do a release, and even then only usually FreeBSD. Besides outside of people like you who intend to use it to troll the BSDs the situation for FreeBSD just really isn't that interesting right now on the open source side, as we know that Radeon is going to perform like crap due to lacking dynamic power management and with Intel he can currently only benchmark up to Ivy Bridge (although the situation is likely to change quite a bit with the release of FreeBSD 11, particularly if linuxkpi makes it in). So as a result the real answer is that it simply isn't worth his time.

Phoronix should conduct more Linux vs. Windows benchmarks. When it comes to Graphics and OpenCL, Linux trails behind Windows. When it comes to Server workloads Linux and Windows are indifferent (which is why Windows' server market share is approx. as high as Linux' ). When it comes to the supervision of "daemons", Linux has this abominable Init-System, er... I mean SystemD deserves an extra chapter in the UNIX-Hater Handbook... What about Windows (NT)? They got it right 20 years ago! If it comes to networking, Linux and Windows are equally bad. That is why MCoreRT*, Seastar**, mTCP*** and OpenFastPath**** have been made.

*( http://www.mcorelab.com/product-rt.html )
**( http://www.seastar-project.org/ )
***( http://shader.kaist.edu/mtcp/ )
****( http://www.openfastpath.org/ )

[starshipeleven]

Phoronix should conduct more Linux vs. Windows benchmarks. When it comes to Graphics and OpenCL, Linux trails behind Windows. When it comes to Server workloads Linux and Windows are indifferent (which is why Windows' server market share is approx. as high as Linux' ). When it comes to the supervision of "daemons", Linux has this abominable Init-System, er... I mean SystemD deserves an extra chapter in the UNIX-Hater Handbook... What about Windows (NT)? They got it right 20 years ago! If it comes to networking, Linux and Windows are equally bad. That is why MCoreRT*, Seastar**, mTCP*** and OpenFastPath**** have been made.

*( http://www.mcorelab.com/product-rt.html )
**( http://www.seastar-project.org/ )
***( http://shader.kaist.edu/mtcp/ )
****( http://www.openfastpath.org/ )

Your post is invalid just because you spelled Systemd wrong.

[aht0]

Why even request benchmarking a code that's running in full debug-mode by default. Try the images, you get to know multiple times during boot that WITNESS is enabled..

Oh, I know requests are motivated by anti-BSD bigotry.. Grow up. You will know how funny that is for people not interested in computers if you come out upstair from your basement and ask your parents

[starshipeleven]

Why even request benchmarking a code that's running in full debug-mode by default. Try the images, you get to know multiple times during boot that WITNESS is enabled..

debug-mode? what is that?

[aht0]

Linux also has "debug" packages. Similar principle. Imagine kernel running like one. You'd have texts like "WITNESS option enabled in kernel, expect reduced performance" in dmesg. Makes sense to have in actively developed code.

[starshipeleven]

Linux also has "debug" packages. Similar principle. Imagine kernel running like one. You'd have texts like "WITNESS option enabled in kernel, expect reduced performance" in dmesg. Makes sense to have in actively developed code.

ah, so the opensource drivers are available only in debug mode. Makes sense, considering their current bad shape.

[Luke_Wolf]

ah, so the opensource drivers are available only in debug mode. Makes sense, considering their current bad shape.

Not what he meant, in releases that are the run-up to a release, software is often built with the debug flag, which means that a bunch of extra code gets compiled in that was hidden behind

#ifdef DEBUG
printf("debug code goes here");
#endif

All that extra code makes things slower, but often it's more than just printing text, there may be expensive state validation code for example. WITNESS as you can read here https://www.freebsd.org/cgi/man.cgi?witness%284%29 validates the state of thread locking in the kernel, which is a very expensive thing to do, and thus will create a serious performance hit.

Additionally if the appropriate flag is passed to the compiler it may include what is known as "Debug Symbols" which allows a debugger to hook into a binary and understand how the machine code corresponds to source code. Debug symbols are different from debug mode, but because of the similarity in naming it's important to bring up to avoid possible confusion.

[starshipeleven]

Not what he meant, in releases that are the run-up to a release, software is often built with the debug flag,

Ah, good point. Alpha 4 should hint at that. SIlly me.

[nasyt]

Oh, I know requests are motivated by anti-BSD bigotry.. Grow up.

Not to forget that all these anti-BSD Guys are Pro-Linux-Bigots. Moreso, they're anti-Solaris, anti-AIX, anti-HP-UX and generally anti-non-Linux. Because, they can't accept, that their Toy-OS sucks!

[TiberiusDuval]

I'd say all OS fanatics be they Windows fanatics, Apple fanatics, Linux fanatics, BSD fanatics, or even few remaining die-hard Amiga fanatics, are equally quite irritating or even toxic. They are blind to failings of their chosen OS, and quite prone to find everything wrong with other systems.